I should probably add that the AW would then be an abstract value that would lead you to choose a particular core tongue width by stack volume to support the flux load. This would be advantageous if you had to design many different OPT's and powers for a wide variety of uses. It would also assume that you had an AW chart that provided a usability rating, the AW number, for practical core stack height vs core tongue width and then added the permeability factor, to arrive at a quick starting point for a design.
This would have been crucial in the 40's and 50's at the various design houses. Perhaps someone with a personal relationship with Mikey at Magnaquest can ask him if any sort of design center charts came along with his data load from Peerless. They would be my first choice for useful short cut charts, just to obtain that all important starting point for a design.
I side stepped the issue by choosing to step up in core tongue width in 1/4 inch increments, just to cut the cost of holding materials minimums for extended periods of time, and also allow me to offer M3 core in some useful size ranges.
Bud
This would have been crucial in the 40's and 50's at the various design houses. Perhaps someone with a personal relationship with Mikey at Magnaquest can ask him if any sort of design center charts came along with his data load from Peerless. They would be my first choice for useful short cut charts, just to obtain that all important starting point for a design.
I side stepped the issue by choosing to step up in core tongue width in 1/4 inch increments, just to cut the cost of holding materials minimums for extended periods of time, and also allow me to offer M3 core in some useful size ranges.
Bud
I just realized why my threat assessment correlator keeps wanting to revisit this thread. This formula has an error in it
Lp = 3,2*A*N²*(1/1/µ+lg/lp))/l*10(power 8)
should read
Lp = 3,2*A*N²*(1/1/µ+lg/lp))/lp*10(power 8)
This scales the formula to the path length. Sorry I didn't see that earlier, but probably no one has been busy designing transformers with it, or you realized that lp was meant in the original formula.
Bud
Lp = 3,2*A*N²*(1/1/µ+lg/lp))/l*10(power 8)
should read
Lp = 3,2*A*N²*(1/1/µ+lg/lp))/lp*10(power 8)
This scales the formula to the path length. Sorry I didn't see that earlier, but probably no one has been busy designing transformers with it, or you realized that lp was meant in the original formula.
Bud
Primary Inductance Questions
There are many companies which sell output transformers. And one of the parameters is inductance (in H) of primary winding. The min required value depends upon plate-to-plate impedance of tubes and lowest frequency. For example, for transformer with 5k primary impedance and 20Hz/-1dB lowest frequency min primary inductance is 80H (159H for 10 Hz).
And here comes the trick. Some manufacturers (e.g. toroidal Plitron/Amplimo) list inductance from 760H (Flow = 27Hz / 4k Raa) till 1056H (specialist range, Flow = 14Hz/-3db, 4k Raa). The primary inductance of these 2 particular units is measured at 60Hz, as stated in product data sheet.
Is there any standard frequency to measure inductance of primary? it is not constant, and changes with signal frequency. And WHY Plitron/Amplimo have SO large primary inductance?
Thanks in advance for any reply.
There are many companies which sell output transformers. And one of the parameters is inductance (in H) of primary winding. The min required value depends upon plate-to-plate impedance of tubes and lowest frequency. For example, for transformer with 5k primary impedance and 20Hz/-1dB lowest frequency min primary inductance is 80H (159H for 10 Hz).
And here comes the trick. Some manufacturers (e.g. toroidal Plitron/Amplimo) list inductance from 760H (Flow = 27Hz / 4k Raa) till 1056H (specialist range, Flow = 14Hz/-3db, 4k Raa). The primary inductance of these 2 particular units is measured at 60Hz, as stated in product data sheet.
Is there any standard frequency to measure inductance of primary? it is not constant, and changes with signal frequency. And WHY Plitron/Amplimo have SO large primary inductance?
Thanks in advance for any reply.
Most manufacturers quote prim L ref at 1Khz; and as you mention the prim L is so high to become meaningless so some manufacturers don't bother specifying it. I agree it's a hopeless figure.
BudP describes a method and if you play with Lg, the final result can give a wild answer from infinity. That's the difference gap/no gap. The predicted gap and approx Fe permeability with an E&I core can be done by re-organising the equation so long the number of turns of known. Physics! Magnetics is often better worked out backwards than forwards.
The reason why Plitron transformers has such a high inductance is due to the continuous Fe strip, which enables near maximum Fe permeability, the cue is the number of turns has to be kept higher than optimum to reduce hysteresis distortion and within flux density. This core with serious out of balance primary currents is easy to saturate.
The standard E&I is in theory a micro-gapped transformer /with laminations up and over each other with oxide films) so the initial Fe permeability is reduced, creating the advantage of a slight straightening and flattening of the BH curve that out of balance primary current is better tolerated but at the lower freq end for the equivalent (compared to toroid) more turns are required,often resulting in higher undesirable interwinding parasitics. The output tranny is a compromise design often wound in several sections.
It is easier to wind compared to a toroid.
Anyone else got an easier way of explaining it !
richy
Not really easier - let alone better,
but I 'grew up' with the 'standard' of measuring primary inductance at 5V, 50 Hz (I believe in the USA that would be 60Hz). The idea is firstly to make figures comparable between products. The practical logic is that (1) inductance rises as the voltage rise, thus this would bring up a relative lowest value. This presumes that the OPT inductance will be a lower pole in the NFB equation, going lower (i.e. going on the safe side) as exitation rises and thus getting itself out of stability factors. (2) The frequency is a moderately low and easily available value, from which extrapolations can be easily made for other values.
This should be a standard, but as Rich said, there are often other conditions used mainly to 'look better' on paper, which is especially irritating when not clearly stated.
This is the logic I became accustomed to.
but I 'grew up' with the 'standard' of measuring primary inductance at 5V, 50 Hz (I believe in the USA that would be 60Hz). The idea is firstly to make figures comparable between products. The practical logic is that (1) inductance rises as the voltage rise, thus this would bring up a relative lowest value. This presumes that the OPT inductance will be a lower pole in the NFB equation, going lower (i.e. going on the safe side) as exitation rises and thus getting itself out of stability factors. (2) The frequency is a moderately low and easily available value, from which extrapolations can be easily made for other values.
This should be a standard, but as Rich said, there are often other conditions used mainly to 'look better' on paper, which is especially irritating when not clearly stated.
This is the logic I became accustomed to.
Amplimo/Plitron's primary inductance measured at 240V 60Hz, as stated in their data sheet.
When you use a commercial inductance bridge, most of them are set to give you a minimum excitation. For US rated machines this is 1 vac @ 120 Hz, or 1 kHz. This provides the minimum inductance, the left hand edge to my perm chart. Plitron uses a maximum excitation point and so their inductance's are the most you can expect. There will still be a perm curve and you will get less inductance as you approach the edges.
Other than this caveat, all that the last few posts have covered is quite correct. There are NO standards for presenting this information, so you must find out the test circumstances, if inductance is important to your end use, and of course, it will be.
Bud
Other than this caveat, all that the last few posts have covered is quite correct. There are NO standards for presenting this information, so you must find out the test circumstances, if inductance is important to your end use, and of course, it will be.
Bud
Ouch.
I learn, but some of the things I learn I do not like.
At least then no international standards - or was I just fortunate in finding firms that did give the 5V-50Hz stuff ....... I got used to using those in preliminary stability calculations (worst case for the OPT), but in the end optimising had to be done in situ (meaning: Build the amp and check stability). This lead to my building a 0,1Hz - 100Hz oscillator to plot down below (darned time-consuming making sure there is no peak there).
Thanks for the enlightenment, Bud!
I learn, but some of the things I learn I do not like.
At least then no international standards - or was I just fortunate in finding firms that did give the 5V-50Hz stuff ....... I got used to using those in preliminary stability calculations (worst case for the OPT), but in the end optimising had to be done in situ (meaning: Build the amp and check stability). This lead to my building a 0,1Hz - 100Hz oscillator to plot down below (darned time-consuming making sure there is no peak there).
Thanks for the enlightenment, Bud!
Johan
On the next set of OPT's, push pull only for now please, stack the core in three chunks. I can provide a PDF document privately if you like showing exactly what I mean, PM me if you wish. Then, before finish varnishing, run the low frequency stability tests, see what you come up with.
Bud
On the next set of OPT's, push pull only for now please, stack the core in three chunks. I can provide a PDF document privately if you like showing exactly what I mean, PM me if you wish. Then, before finish varnishing, run the low frequency stability tests, see what you come up with.
Bud
I was but ...
I am interested in ALL METHODOLOGIES that people use -- please share yours!!
I have learned alot (and continue to) and received great advise from this post so far. Most importantly about all the things I needed to learn which I am in process of (better than watchin the boob tube
). My wife and I discuss (and laugh about) this post/forum all the time. The ecclectic nature of this post is JUST what I wanted -- thx to all. Keep it coming
I am interested in ALL METHODOLOGIES that people use -- please share yours!!
I have learned alot (and continue to) and received great advise from this post so far. Most importantly about all the things I needed to learn which I am in process of (better than watchin the boob tube
). My wife and I discuss (and laugh about) this post/forum all the time. The ecclectic nature of this post is JUST what I wanted -- thx to all. Keep it coming
Last edited:
Yup, exactly that. However....... life is usually much more complex than a quickie thumbnail riule allows for and transformers are no exception.
If you want your OPT to actually match your tube, across the frequency response band and you want to be the person who determines what to buy, what Johan and I have portrayed is what the rule of thumb must contain. Otherwise, you have to go to a designer, like Johan, I, Per Lundahl, Dave Slagle etc, or the fine folks at Hammond, Edcor, James, etc and have them do the work for you.
The Hammond catalog specs will get you close enough that your amp will work and you will recognize the sounds being played through it. That is as close as a rule of thumb choice can get you, and there is NOTHING wrong with this approach. For prototyping and casual building for proof of concept, I recommend that do go through Hammond or other generalist manufacturers. For something more revealing, the devil is in the details and you have to know about them and be ready to use them.
Bud
Same from me, Bud - you obviously know your magnetic material' behaviour.
A question then from me regarding the frequency response of core material. My take is that basically this depends on the thickness of the laminations - the thinner the better. But these days where just about every winder is into power transformers (50Hz or 60Hz), and those doing disco equipment output transformer rewinding not really wiser, how do I tell the h.f. capabilities of a steel? You might have picked up that for hi-fi stuff I prefer using C-cores (thin sheet); are they otherwise the same?
Some info in this regard will be appreciated (or a web-site I overlooked).
Thanks
A question then from me regarding the frequency response of core material. My take is that basically this depends on the thickness of the laminations - the thinner the better. But these days where just about every winder is into power transformers (50Hz or 60Hz), and those doing disco equipment output transformer rewinding not really wiser, how do I tell the h.f. capabilities of a steel? You might have picked up that for hi-fi stuff I prefer using C-cores (thin sheet); are they otherwise the same?
Some info in this regard will be appreciated (or a web-site I overlooked).
Thanks
Hi Johan,
Well, you are quite knowledgeable yourself. I only have some postulates to pass on, concerning core material and high frequencies.
1.) The thinner the core material the finer grained will be the planar magnetic fields, constructed within the window of E/I core. I think this is true of wound core also, but the layers of core are not edge laid into the window with C core etc. Also, I have never found a C core with thicker than 0.006 material so I really have no way to compare the theoretical mag field structures between the two. It is entirely possible that C core is superior here, with only two very large fields, rather than many very thin ones.
2.) Commercial E/I core materials are not in the picture, for being the transformation vehicle of signals, much above 400 Hz. Anything thicker than 29 gauge M6 is useless by 250 Hz. They still act as a ferrous bounding box. They still dictate the speed with which a magnetic field can be created within the window, and thus potential signal rise time, but a direct antenna to antenna event is the true transformation activity above these frequencies. Again my study of C Core in this respect is very limited, but I have not found any reason to think that power loss curves for wound core differ from E/I core materials. I am still discussing non nickle iron core materials here.
3.) My findings on core material frequency response limitations, or, just how far up the frequency response curve core is causing distortion in the transformed signal, are as follows. In a general sense M50 to M19 250 Hz, M6 & M3 400 Hz, 48% nickle iron 3500 Hz, 80% nickle Iron 7500 Hz and Amorphous core 18,000 Hz. Again, these come from power efficiency curves. All core will respond to all frequencies applied to it, but the effective square area of usable flux drops at a faster rate than the square area required by the frequency in question and so, power transformation from primary coil through core to secondary coil, is negligible above those general numbers expressed above.
4.)What I glean from this is that you want to study antenna theory and the dielectric composite number called dielectric constant. In short, if you use commercial core you must utilize antenna shadowing characteristics and very tight capacitive coupling, so as to make the best use of both E Field and B Field activities. If you use Amorphous core you still want to utilize antenna shadows, but you want the least possible capacitive coupling. You can see these two extremes at work in my audio transformers and those of my esteemed colleague Per Lundahl. Working somewhere in between these two poles requires a delicate balancing act, when it comes to choice of how much capacitive coupling to use and what sort of dielectric constant characteristics you want to incorporate. Two very good practitioners of this balance are Pieter Truffant of Tribute Audio and Dave Slagle of Intact Audio. Which route you take to high performance is really based more upon taste than any large performance differences, once you have understood how to make commercial E/I core work properly for audio concerns. I can assure you that in any case, power transformer E/I construction is the antithesis of what you want for Audio.
5.) All gap structures are your enemy and you want to push them as far away from the coil as possible and have them be uniform, but not contiguous through a stack of core material, in close proximity to a coil.
6.) So, in a general sense, what you do with your coil is as important as what you do with your core. You must match the relative permittivity (for want of a more useful term) of both your magnetic and dielectric circuits, to achieve high resolution, transient faithfulness, and overall low distortion.
Bud
Well, you are quite knowledgeable yourself. I only have some postulates to pass on, concerning core material and high frequencies.
1.) The thinner the core material the finer grained will be the planar magnetic fields, constructed within the window of E/I core. I think this is true of wound core also, but the layers of core are not edge laid into the window with C core etc. Also, I have never found a C core with thicker than 0.006 material so I really have no way to compare the theoretical mag field structures between the two. It is entirely possible that C core is superior here, with only two very large fields, rather than many very thin ones.
2.) Commercial E/I core materials are not in the picture, for being the transformation vehicle of signals, much above 400 Hz. Anything thicker than 29 gauge M6 is useless by 250 Hz. They still act as a ferrous bounding box. They still dictate the speed with which a magnetic field can be created within the window, and thus potential signal rise time, but a direct antenna to antenna event is the true transformation activity above these frequencies. Again my study of C Core in this respect is very limited, but I have not found any reason to think that power loss curves for wound core differ from E/I core materials. I am still discussing non nickle iron core materials here.
3.) My findings on core material frequency response limitations, or, just how far up the frequency response curve core is causing distortion in the transformed signal, are as follows. In a general sense M50 to M19 250 Hz, M6 & M3 400 Hz, 48% nickle iron 3500 Hz, 80% nickle Iron 7500 Hz and Amorphous core 18,000 Hz. Again, these come from power efficiency curves. All core will respond to all frequencies applied to it, but the effective square area of usable flux drops at a faster rate than the square area required by the frequency in question and so, power transformation from primary coil through core to secondary coil, is negligible above those general numbers expressed above.
4.)What I glean from this is that you want to study antenna theory and the dielectric composite number called dielectric constant. In short, if you use commercial core you must utilize antenna shadowing characteristics and very tight capacitive coupling, so as to make the best use of both E Field and B Field activities. If you use Amorphous core you still want to utilize antenna shadows, but you want the least possible capacitive coupling. You can see these two extremes at work in my audio transformers and those of my esteemed colleague Per Lundahl. Working somewhere in between these two poles requires a delicate balancing act, when it comes to choice of how much capacitive coupling to use and what sort of dielectric constant characteristics you want to incorporate. Two very good practitioners of this balance are Pieter Truffant of Tribute Audio and Dave Slagle of Intact Audio. Which route you take to high performance is really based more upon taste than any large performance differences, once you have understood how to make commercial E/I core work properly for audio concerns. I can assure you that in any case, power transformer E/I construction is the antithesis of what you want for Audio.
5.) All gap structures are your enemy and you want to push them as far away from the coil as possible and have them be uniform, but not contiguous through a stack of core material, in close proximity to a coil.
6.) So, in a general sense, what you do with your coil is as important as what you do with your core. You must match the relative permittivity (for want of a more useful term) of both your magnetic and dielectric circuits, to achieve high resolution, transient faithfulness, and overall low distortion.
Bud
Last edited:
Even very old (manufactured in 50th - 60th) high-quality 0.3-0.5 mm silicon steel lamination works fine in audio frequency transformers. Output transformers in vintage Pioneer SM-83 (manufactured by Tamura at the end of 60th) have virtually flat response up to 50 KHz (Pioneer claims 100 KHz but my digital input interface ESI Juli do not allow to measure over 50 KHz).
Taking surplus cores doesn't make any sense (at least for me) unless you precisely know what kind of material it is and what are its characteristics. This is not only permeability f(flux curve), but also permeability curve f(DC current over coil, flux) and "a"/"am" coefficients f(flux, DC current over coil), etc. I'm in doubt you can find this kind of data for material taken from scrapyard. In fact, you can barely find even for today's ferromagnetics. H-Fi audio-frequency transformers market is tiny, so no metallurgy factories takes real care about it. They have a number of alloys in their product line, which may work great, or may not, its all up to your skills.
Without CAD-based simulating of all this data + coil/winding geometry + amplifier output stage characteristics, constructing good audio frequency transformer have the almost same possibility as winning big bucks in lottery (although I cannot exclude there are talents who can). I have a software (BA TrafoCalc CAD) which upon evaluating all these input parameters can placate a poorly designed transformer, which, under certain circumstances, can produce as much as 3% of nasty 3rd order harmonic distortions, therefore, turning any amplifier into junk box.
Last edited:
I was with you until you mentioned price, nonetheless you might want to talk to Jack Elliano at Electra-Print about your needs - he has wound a number of custom OPTs for me and I can honestly say I have been completely delighted with the results.
Jack might be able to build you something for not too much more than your target price point.
Electra-Print.com Audio Transformers
No affiliation, just a happy customer.
Edcore is another one to take a look at - many here swear by their products:
EDCOR Electronics Corporation
Note that high nickel content or amorphous cores will drive the cost way up from where you say you want to be. (My last Japanese made amorphous cores were over $600 each) For your budget you can get M6 which should be just fine in the right hands...
Last edited:
I agree with Kevin, Edcor and Hammond are quite good. Not cutting edge in information retention but certainly as good as anything from the commercial realm in the 50's. You can PM me, but I will be about 50% above your target price, the amount of retained data will be worth the difference, if you have a set of speakers that will reveal it, most mid fi and up will do so.
Bud
Bud
Bud, I don't want to PM without permission but have a question regarding your transformers. I have been reading and enjoy your detail, but my high energy physics background is way rusty. On a practical side, you previously mentioned some sites, such as Rhodes as having transformers I assume are your design. Are these the Onetics? Are they suitable for an SE 300b design as offered on these sites or are you recommending something else for thor above? If so, can I PM you about options? I have a few possibilities lined up, but I am interested in what you might say and how they might compare. Although I should be satisfied with simpler, less expensive parts, I can't. I like good stuff, of course within some reason. Thanks. Sorry for the hijack.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.